How to Remove Backgrounds From Images with Bria RMBG 2.0

Add an Image node and upload the photo you want to cut out. The model works on any image type: product photos on white or cluttered backgrounds, portraits with complex hair, logos on colored backgrounds, even illustrations. Higher-resolution source images produce cleaner edge detail — if you have both a web-size and full-resolution version, use the full-resolution one. You can also connect the output of another Image node (e.g., an AI-generated product shot from FLUX.2) directly as input.

Add a Tool node to your canvas and select "Bria RMBG 2.0" from the tool list. Draw a connection from your Image node to the Tool node input. There are no parameters to configure — no quality settings, no edge refinement sliders, no output format choices. Bria RMBG always outputs a transparent PNG at the same resolution as the input image. This zero-configuration design means the result quality depends entirely on the input image quality, not on any settings you might miss.

Click generate. The cutout appears quickly. For most images, the edges are production-ready immediately. Pay special attention to three challenging categories: (1) Fine hair and fur — Bria preserves individual strands rather than creating a hard crop, but stray wispy hair on very busy backgrounds may get clipped. (2) Semi-transparent objects — glass, sheer fabric, and smoke are partially preserved with appropriate transparency levels. (3) Overlapping subjects — when a hand or product extends beyond the main body, check that the full silhouette is preserved.

The cutout on its own is useful for design tools, but the most powerful workflow chains Bria RMBG with an image generation model. Connect the transparent PNG output to a FLUX Kontext node and prompt: "Place this product on a minimalist white marble surface with soft studio lighting." The Kontext model receives the cutout as a reference and composites it onto the described scene with natural shadows and reflections. This two-node pipeline (RMBG → Kontext) replaces what traditionally requires a photographer, a studio, and a designer.